CITIES AS LABOR MARKETS - Marron Institute · 2019-07-11 · CITIES AS LABOR MARKETS + ALAIN BERTAUD ABSTRACT A city’s welfare depends on its labor market. As long as a labor market

CONTACT

Alain Bertaud [email protected]:urbanizationproject.org

WORKING PAPER #2 / FEBRUARY 19, 2014

CITIES AS LABOR MARKETS

+ ALAIN BERTAUD

ABSTRACT

A city’s welfare depends on its labor market. As long as a labor market does not fragment into adjacent,

smaller ones as it grows, the larger the market, the more innovative and productive the city will be.

Maintaining mobility is therefore essential to the economic viability of cities.

Maintaining mobility has two implications: first, managing a transport system that allows an efficient

movement of labor and goods across metropolitan areas, and second, insuring that regulations or inadequate

land supply do not prevent firms and households from settling in the area that will maximize their welfare.

This also implies that transaction costs should be low enough to allow firms and households to change

location when their circumstances change. Considering cities primarily as labor markets has important

operational implications for their management, particularly in the way transport systems are developed and

in the way land markets are allowed to operate.

THE EFFICIENCY OF LARGE LABOR MARKETS IS

THE MAIN CAUSE OF EVER-GROWING CITIES

c i t i e s a r e p r i m a r i ly l a b o r m a r k e t s

Cities are primarily labor markets. This claim may seem terribly

reductionist to the many among us who love cities. Certainly

the attractions offered by the amenities of a large city cannot be

reduced such that the whole is seen merely as a place where firms

are looking for labor and people are looking for jobs.

During the French “cultural revolution” of May 1968 students

were deriding a life reduced to only three activities: “Metro,

boulot, dodo”, which could be roughly translated by “commuting,

working, sleeping.” This became one of the most ubiquitous tags

on Paris’ walls. The students were revolting against what I, more

pedantically, call an urban labor market, and they had a strong

point; I have seen such reduced forms of urban life in many

malfunctioning cities. However, I believe that improving the way

labor markets function through better land use and transport

allow for the indispensable values of urban life: a commute short

enough that one has time for elective activities; an open job

market that allows one to change jobs until an interesting and/

or materially rewarding professional activity has been secured; a

residence from which access to social life or nature is quick and

easy. All this is possible only in the presence of a well-functioning

labor market. Urban life reduced to “Metro, boulot, dodo” is

precisely the expression of a dysfunctional labor market.

I am not implying that a city’s only purpose is as a labor

marketplace, but I am arguing that without a functioning labor

market there is no city. Try thinking of an alternative explanation

for the existence of very large cities. A city nucleus might have

been created originally as a commercial port, a trading post,

an administrative center, a military stronghold, or a center

of religious pilgrimage, but over the years, the growth of a

diversified labor force would be the only possible cause for the

expansion of the original urban nucleus. While most cities offer a

lot more than job opportunities, it is important to recognize that

the expansion of job markets makes everything else possible. A

well-functioning labor market brings together people with varied,

but complementary, knowledge and skills-- the preconditions

for innovation. A well-functioning labor market makes possible

every other urban attraction—symphonic orchestra, museums,

art galleries, public libraries, well-designed public spaces, and

great restaurants, among many others. In turn, these typically

urban amenities require additional specialized jobs and attract

an even more diverse population, which become the source of

future innovations and a more interesting urban life.

Usually when a city’s population is growing, it means that its

labor market is growing. Still, there is a segment of any urban

population (usually between 35 and 50 percent) that does not

participate directly in the labor market. Statisticians rightly

call the non-active segment the “dependent” population. Its

members—retired people, infants, students, prison inmates,

etc.—are not part of the labor market and participate in the urban

economy only as consumers.

People migrating from other cities once they have reached

retirement age may be the cause of the growth of a few cities

whose growth is more driven by consumer markets than by labor

markets. These types of cities might become more common in

the twenty first century with the projected aging of the world

population. The retiree population of these cities would be

expected to consume a lot of services in health care facilities,

restaurants, and entertainment venues. The growth of these

“retiree” cities would then be caused by the dual effect of not

only the retirees’ migration, but also the migration of additional

workers to staff the services required by the retirees. These

retiree cities would not require spatial concentration and are

unlikely to create much economic dynamism. The eventual

growth of retiree cities is the only exception to growth created

by the efficiency of large labor markets. And, of course, the

retirement income of retirees will have to have been generated by

efficiently working labor markets in other large cities.

l a r g e l a b o r m a r k e t s a r e m o r e p r o d u c t i v e t h a n

s m a l l e r o n e s

Economists have convincingly demonstrated the productivity

advantage of larger cities over smaller ones. Large cities generate

scale economies that allow enterprises to reduce their costs

by increasing output, thereby reducing costs per unit. Scale

economies are only possible in cities with a large labor market.

When many related activities are located in close proximity, they

generate what economists call “knowledge spillovers.” New ways

of doing things in one firm are soon imitated by other firms and


eventually by other sectors as a result of the proximity and close

contact between workers of different firms and sectors within

the urban economy. For instance, the first users of electronic

spreadsheets in the early eighties were mostly accountants

and financial analysts. The use of spreadsheets soon became

common in all sectors of the economy, but the spillover occurred

first in large cities. Knowledge spillovers are responsible for

agglomeration economies, i.e. increase in productivity due to the

rapid dissemination of new ideas in areas where large numbers

of workers are in close contact.1 Agglomeration economies also

result from a lowering of transaction costs in larger cities because

of the proximity of competing suppliers and consumers.

Economic literature linking the wealth of cities to spatial

concentration is quite abundant and is no longer controversial

in academic circles. National accounts show that the output

share of large cities is always much higher than their share of

the national population. The 2009 World Bank Development

Report “Reshaping Economic Geography,” and the report of the

Commission on Growth and Development, “Urbanization and

Growth” published the same year, exhaustively summarize and

document the theoretical and empirical arguments justifying

the economic advantage provided by the spatial concentration of

economic activities in large cities.

But if larger cities are more productive than smaller ones, why

are large cities not growing faster than small ones? And why do

many households and firms choose to remain in or even move

to smaller cities when they could instead settle in the more

productive environment provided by larger cities?

t h e p r o p o r t i o n o f s m a l l e r c i t i e s to l a r g e r o n e s

s tays co n s ta n t ov e r t i m e; o n av e r ag e , a l l g r ow at

a b o u t t h e s a m e r at e

Data on city size distribution by country or by region show that

the proportion of small to medium and large cities stays about

constant over time. When households decide to migrate and

firms decide to select a location for a new enterprise, they are as

likely to choose a small city as a larger one.

The American economist Vernon Henderson, who pioneered

work on the growth rate and size distribution of cities in various

1 The internet is certainly able to spread knowledge quickly without requiring spatial concentration. However, the

impact of the internet in disseminating knowledge might be similar to books: it makes knowledge available quickly

and cheaply but it doesn’t replace the serendipity of a random meeting of people with similar interests.

countries, shows the regularities found in the distribution of

city size across countries, with the exception of anomalies in

the former Soviet Union and China. In his book Planet of Cities

(2012), Shlomo Angel summarizes previous studies on the subject

and addresses the issue of worldwide city size distribution.

Angel based his analysis on a reliable worldwide database. His

conclusions confirm previous, less exhaustive studies:

• Thesizedistributionofcitiesbycontinentfollows

Zipf’s law, which states that the size of a city’s

population is inversely proportional to its statistical

rank, such that if cities are ranked by decreasing

size, the second largest city is half the size of the

largest, the third largest city a third the size of the

largest, etc.

• Onaverage,largecities,asagroup,aregrowingat

about the same rate as medium and small cities in

the same countries or regions. It seems that cities’

growth rates follow Gibrat’s law of proportionate

effect, which says that the size of a city is not an

indicator of its future growth rate—that is, cities’

growth rates are random, with the same average

expected growth rate and same variance. This is why

Zipf’s law is preserved over time.

In any given region, the distribution of cities of various sizes

therefore remains stable. Larger cities keep growing, but on

average, so do smaller cities. This seems paradoxical given that

larger cities are more productive than smaller ones. However,

larger cities do not play the same economic role as smaller

cities. They complement each other’s activities. The increased

productivity of larger cities is therefore linked to the existence

and growth of smaller cities. In turn, smaller cities’ economic

growth is dependent on larger cities’ innovations and inventions.

s o m e c i t i e s k e e p g r ow i n g w h i l e ot h e r s d o n ’t

Small cities do not always grow into larger cities. The rate of

population growth is determined by economic opportunity,

which in turn is largely determined by the comparative advantage

given by a city’s location and its population’s capacity for

innovation. But the economic advantage provided by location


is not necessarily permanent; it may increase, decrease, or even

vanish with technological change. Being close to an obsidian

mine might have been a decisive advantage for the early Middle

Eastern cities described by Jane Jacobs, but that advantage

disappeared when obsidian ceased to be the preferred material

for tools and weapons. The Anatolian cities, whose economies

had not been able to diversify into activities other than obsidian’s

craft and trade, inevitably shrunk and eventually disappeared.

The dominance of New York as the United States’ main eastern

seaport was made possible by the comparative advantage

provided by the Erie Canal. By the time railways made waterway

transport obsolete, New York’s population had accumulated such

a high level of diversified, specialized skills that it continued

to thrive without having to rely on the advantage conferred by

proximity to the canal.

The history of the world’s cities is full of examples of large cities

dominating their regions for a time and then shrinking back to a

smaller size or even into oblivion. In 1050, Cordoba, in the south

of Spain, was the largest city in Europe with 450,000 inhabitants,

followed by Palermo, Sicily, with a population of 350,000. By the

middle of the fourteenth century, the population of both cities

had shrunk to 60,000 and 50,000, respectively. In the eleventh

century, Kaifeng in China was probably the largest city in the

world, with 700,000 people, while Shenzhen was not even on

the map. Today, Shenzhen has 10 million people, and over the

past ten centuries, Kaifeng’s population has barely increased to

800,000 people.

w h y d o n ’t h o u s e h o l d s a n d f i r m s m i g r at e to l a r g e r

c i t i e s w h e r e p r o d u c t i v i t y a n d s a l a r i e s a r e h i g h e r?

In spite of the higher productivity to be gained, only some types of

firms can benefit from moving to a larger city. Firms established

in larger cities require higher capital and higher operation and

maintenance costs than those located in smaller cities. Land and

rents are more expensive in larger cities than in smaller ones.

Distances traveled are longer, and the “congestion tax” is higher.

In addition, not every enterprise can benefit from economy of

scale or agglomeration economies.

Moving to a smaller city where land is cheaper and salaries are

lower makes economic sense for firms whose activities require

a lot of land and labor that is not particularly specialized. For

instance, activities like furniture-making require a lot of land and

easy truck access to transport the bulky materials necessary for

the finished product. They require skilled, but not particularly

specialized, labor. Furniture manufacturers, therefore, have no

reason to relocate to a large city, where land and labor would be

expensive and where moving bulky raw materials and finished

products in and out of the factory would be inefficient and costly.

These types of firms would tend to locate their manufacturing

activities in smaller cities. However, furniture makers may

require innovative designers who may not be found in smaller

cities. In such cases, they may need to subcontract the design

of furniture to a firm located in a large city, where talented

designers are more likely to be found and where agglomeration

economies and idea spillover, both important for a design

firm, are also more likely to occur. Firms such as furniture

makers can carry out their highly specialized and innovative

activities—design and marketing, for instance—in large cities.

Their repetitive and land intensive activities (manufacturing) can

be carried out in smaller cities. In this way, they can enjoy the

advantages of both a large city (innovation, specialized labor) and

a smaller city (low land and labor costs).

Speedier and cheaper communication over the past twenty years,

including the expansive adoption of the Internet, has contributed

to the splitting up of large firms into various departments located

in cities of different sizes. Specialized tasks (design, marketing,

export promotion) can take place in larger cities, where the

requisite innovators and specialized labor force are more likely

to be found, while more routine manufacturing can take place in

smaller cities. In addition, large firms increasingly subcontract

tasks to smaller firms often located in different areas. The same

factors have likely led to the growth of both large and small

cities, allowing them to specialize in what they do best. Similar

rationale could be applied by workers who prefer either to remain

in large cities or to migrate to smaller cities: the latter have lower

salaries but also lower rents, lower commuting costs and, often, a

better natural environment.

Some services are likely to thrive in both large and small cities

and are thus not dependent upon the advantage provided by

location. Fast food restaurants, barber shops, and laundry

services, for instance, follow the labor force of more specialized

firms wherever they locate, contributing to the even growth rate


of small and larger cities.

THE PLANNER’S ANTI-BIG-CITY BIAS AND

THEIR ATTEMPTS TO “BALANCE GROWTH”

Cities grow when their labor markets expand. This economic

expansion is usually the result of a comparative advantage gained

from location or an unusual concentration of skilled workers.

The rate of a city’s population growth cannot be attributed to

advance planning; rather, it is due to a combination of exogenous

and endogenous circumstances. To the chagrin of urban

planners, a city’s growth rate over the mid or long term is largely

unpredictable, and it is futile to pretend it is the result of careful

planning.

Planners and city managers have traditionally been concerned

about the unplanned growth of large cities because of the

complexity involved in managing them, the difficulty integrating

poor migrants from rural areas into city life, and an instinctive

aversion to anything that seems “undesigned.” Planners have

even described the growth of large, dominant cities like Paris or

Mexico City as “cancerous.”

The aversion to unplanned or to “asymmetrical” spatial patterns

is quite apparent in most urban planners’ approach. Some

planners look at a country’s map and observe that some regions

contain many cities while others have only a few. They incorrectly

conclude that this “imbalance” represents an inequity due to

parasitic urban activities or to other market failures. In their

view, it then becomes the responsibility of the government to

modify the imbalance and to remove this regional inequity

through national spatial planning, with the declared objective of

restoring a regional symmetry in the spatial distribution of cities.

However, the assumption that national spatial planning can

modify the distribution of urban populations in order to reach a

new, planner-designed spatial equilibrium is false.

Cities that have a decisive comparative advantage, either

because of their location or because of their large specialized

and innovative labor pool, are likely to grow. People migrate

toward cities where economic and social opportunities are best,

from their point of view. The idea that a city’s economic and

demographic growth rate is due to parasitic activities occurring to

the detriment of other cities is fanciful—unless, of course, piracy,

smuggling, or other unlawful, predatory activities are its main

cause.

The assumption that the preparation of national or regional

plans would result in a predictable urban growth rate for

each individual city in a region is also demonstrably false.

Unfortunately, in many countries this common planning conceit

has resulted in misallocated public investments and regulatory

impediments that have decreased cities’ productivity. In reality,

planners have very little influence on city size distribution and

city growth rates, unless they take active, targeted measures to

destroy the urban economies of the cities that have grown “too

large.” The Khmer Rouge’s urban policy applied to Cambodia in

the late seventies was an extreme and brutal example of planners’

temporarily successful attempt to manage city size.

As a consequence of the planners’ hubris about the necessity

of managing city size, many regional plans designed in the

second half of the twentieth century have promoted regulatory

limits on the growth of large cities. These were combined with

planned infrastructure investment aimed at stimulating the

growth of smaller cities, which were deemed more manageable.

A seminal and influential paper published in 1947, calling for a

national plan for the spatial development of France, was titled

“Paris and the French Desert,” implying that the growth of Paris

had occurred at the expense of the French provincial towns.

Anyone familiar with French provincial towns would recognize

their comparison to a barren desert as a slightly comical but

gross exaggeration. While it is possible that the centralization

tendency of successive governments since the French revolution

of 1789 contributed to Paris’ rapid growth, the problem, if it

exists, lays with the political system. Preventing investments in

the capital while directing large resources towards provincial

towns is unlikely to change a city size hierarchy caused by an

idiosyncratic political system whose reforms fail to allow for more

decentralized decisions.

In 1956, the Indian government adopted a policy dictating that

new industries should locate in “backward areas.” At the same

time, it prevented further development of manufacturing in

large cities.2 Through this policy, the government committed

2 Industrial Policy Resolution of the Government of India adopted in 1956 under the provisions of the Industrial

Development and Regulation Act, 1951.


itself to correcting regional imbalance and to preventing further

industrial growth in cities of more than 500,000. In 1988, the

negative impact of the policy was compounded by an interdiction

that new industries locate less than 50 kilometers from cities with

a population of more than 2.5 million and within 30 kilometers

of cities with a population between 1.5 million and 2.5 million. As

one can easily imagine, the latter policy didn’t prevent the growth

of industries in successful cities like Mumbai or Bangalore with

a population significantly bigger than 2.5 million; it just made

it more expensive for these industries to expand there. More

tragically, it diverted scarce government infrastructure resources

to regions with weak potential while starving large metropolitan

areas of desperately needed investment, even though this

was where most people were migrating. The current poor

performance of public infrastructure – roads, transport, sewer,

drainage, and power – in major Indian cities is in part the result

of misguided national spatial policy conducted over the last 50

years.

If planners are unable to control the growth rate of cities, how do

I explain the successful growth of entirely planned cities like St

Petersburg, Brasilia or Shenzhen, created ex nihilo by powerful

rulers as diverse as Peter the Great, Juscelino Kubitschek and

Deng Xiao Ping? These planned cities became large and

successful as the result of two main factors. First, each city’s

location was selected because of a geopolitical necessity3 and

not because of an abstract planning concept. Second, each city

had the strong political and financial support of a powerful ruler

of a very large country. This support allowed these cities to sink

large amounts of money in infrastructure investment without

having to borrow and tax their own initially fledgling economies.

These political and financial conditions are not typically met in

the regional plans routinely prepared by technocrats. Politicians

created new capital cities like Washington (USA), Canberra

(Australia), Islamabad (Pakistan), Abuja (Nigeria), and Naypyidaw

(Myanmar). All are capitals of large countries and were initially

without economic base beyond the national government

bureaucracy. The “cost is no object” concept presided over

3 St. Petersburg was created by Peter the Great to open a port toward Western Europe in order to gain new technology

through trade and cultural contact. Brasilia, created by president Juscelino Kubitschek of Brazil, was part of an

effort to develop the center of the country and to make the capital more politically independent from the large cities

of the coast. Deng Xiao Ping’s main objective in creating Shenzhen was to graft and test within a limited perimeter

some of the market institutions and technical know-how used across the border by his Chinese compatriots in Hong

Kong.

their construction and insured their initial survival as they were

financed by taxes paid by the rest of the country. Eventually, a

more diversified labor market grafted itself onto the government

activities.

During the 70 years of the Soviet Union, planners had the

opportunity to decide which cities were going to grow and which

were not. No city could grow without supporting resources

allocated from the Gosplan,4 a specialized ministry in Moscow.

The government had the means to enforce the movement of

people, and migrations toward selected locations within the

Soviet Union’s vast hinterland were often involuntary. Many new

cities were created for various political or perceived economic

reasons, but none of these cities was the result of voluntary

migration of firms and people toward areas that represented

better opportunities.

In 2010, traveling to Moscow as a consultant, I was asked by the

ministry of construction to provide advice on how to proceed

for the “closing” of 60 cities that the Russian government had

identified as no longer viable. The government could not

continue to support social services and infrastructure in cities

that had been abandoned by the large monopolistic industries

that were originally their raison d’être. The labor market had

disappeared, but the laborers were still there; “closing” the

cities would entail another forced migration of several million

people. Apartments that had recently privatized represented

most people’s only asset; however, because the apartments

had become worthless, their owners were unable to move. The

closing of cities in Russia is an extreme illustration of the danger

of creating cities based on “planning” criteria without economic

base and of using forced migrations or heavy subsidies to

promote urban growth.

w h y p l a n n e r s s h o u l d n ot t ry to a lt e r t h e

d i s t r i b u t i o n o f c i t y s ize s

There is a “natural” equilibrium reached, within countries

and regions, between the size of the population and the firms

choosing to settle in small, medium, and large cities. This

equilibrium is created by the accumulated decisions of firms and

households to “vote with their feet,” thereby selecting to move to

4 In Russian: ”Gosudarstvennaya Planovaya Comissiya” State Planning Committee, in charge of the Soviet

economy.


the cities that will grow and to leave the cities or villages that have

less potential. The spontaneous spatial equilibrium created by

the sum of uncoordinated, individual decisions illustrates what I

call the principle of “order without design.”

With the exception of the few geopolitical examples mentioned

above, planners have no credible rationale for intervening directly

in the location and growth rate of cities. Planners should no

more “encourage”—a favorite word in the planning literature—

the growth of large cities at the expense of smaller cities than

they should discourage their growth, as they have done in the

past. History has shown that these types of planner initiatives

are bound to fail or, worse, to create serious diseconomies,

making a country poorer. The size of a city does not make it

automatically more productive—large, dense refugee camps are

less productive than small towns, although they may provide

shelter to several hundred thousand inhabitants. A city only

becomes more productive when its growth is generated by the

aggregated decisions of many firms and households to migrate

toward it and when those firms and households have the freedom

to either stay put or to migrate elsewhere. These decisions should

not be altered either by coercive regulations or by government

investment incentives. Because households and firms have the

most invested in the successful outcomes of their moves, we

have to trust that the majority of them have enough information

to justify their migration choices. Planners, in contrast, lack

the information about the economy of individual firms and

households that would be necessary to make informed decisions

about the advantages and disadvantages of locating in a small,

medium, or large city.

We will see below that planners’ “optimum design” hubris is

not limited to the size and location of cities. Within cities,

too, they attempt to regulate both where households and firms

should locate and the quantity of land and floor space they

should consume. As we will also see, planners do have a crucial

role to play in the development of cities, in particular with the

development of their infrastructure. However, it must be clear

that allocating land and floor space in specific locations is not

their role.

THE CITY’S PRODUCTIVITY DEPENDS ON ITS

ABILITY TO MAINTAIN MOBILITY AS ITS BUILT-

UP AREA IS GROWING

Mobility, which I would argue is the centerpiece of

our national productivity, is neither highly valued

nor understood among public officials.

--Alan Pisarski, 2006

Good management can therefore increase

indefinitely the «optimal» size of a city.

--Rémy Prud’homme & Chang-Woon Lee, 1998 ,

“Size,Sprawl,SpeedAndTheEfficiencyOfCities”

Larger labor markets are made possible by an increase in the

mobility of people and goods. Advancements in urban transport

technology have improved the mobility of people and goods,

which in turn, has contributed to the growth of large cities during

the last 150 years. Improvements to transport technology have

also made possible the spatial concentration of both people and

fixed capital. Economists describe fixed capital as factories,

office buildings, houses, apartment buildings, community

facilities, and infrastructure. In the last fifty years, an increasing

return to scale and agglomeration economies as a result of this

spatial concentration has led to the emergence of mega-cities.

The potential economic advantages of large cities are reaped only

if workers, consumers, and suppliers are able to exchange labor,

goods, and ideas with minimum friction and to multiply face-to-

face contacts with minimum time commitments and cost. The

productivity of a city with a growing population can increase only

if travel between residential areas and firms and among firms’

locations remains fast and cheap. As a city grows, it is therefore

important to monitor mobility by comparing how average travel

times and transport costs vary over time.

t h e da i ly h u m a n t i d e: t h e c h a l l e n g e o f m ov i n g p eo p l e

a n d g o o d s

The necessity of managing urban growth rather than to trying

to slow it down is finally being understood by mayors, city

managers, and urban planners. An increase in city size is not the

only condition necessary to increase productivity. Productivity

increases with city size only if the transportation network is able


to connect workers with firms and providers of goods and services

with consumers. This connectivity is difficult to achieve in large

cities; it requires consistency among a number of factors: land

use and investments for transport networks, pricing decisions

for road use, parking, and transit fares, and collection of local

taxes and user fees. In their 2009 book, aptly titled “Mobility

First,” Sam Staley and Adrian Moore describe in detail the cross-

disciplinary reforms in road and urban transport design and

in road pricing, among other things, that would be required to

maintain mobility in cities in the twenty first century.

Failure to manage urban transportation in a manner that

maintains mobility results in congestion. Congestion decreases

labor mobility and productivity and is in fact avoidable in

large cities. Its presence represents a failure on the part of city

managers. Congestion has a double-negative effect; it acts

as a tax on productivity by tying down people and goods, and

it degrades the environment and increases greenhouse gas

emissions. It is conceivable that in the future some mismanaged

large cities may reach a level of congestion and pollution the

combined negative effects of which could offset the economic

advantage of spatial concentration. These cities would then stop

growing, and the economic advantage of spatial concentration

would be taxed away by congestion and an unsafe environment.

Given this potential scenario, the positive economic effect of

agglomeration must be very powerful in cities like Bangkok and

Jakarta where urban productivity continues to offset the price of

chronic congestion. It is difficult to assess a city’s productivity

just by visiting, but traffic congestion is clearly apparent on even

a short visit to either city. But, even the spectacular and semi-

permanent traffic congestion in these economic powerhouses, as

well as in Beijing, does not cancel the productivity advantage of

their large, human spatial concentrations.

Maintaining mobility while a city’s built-up area and its

population are growing is not easy. During centuries of urban

development, walking was an adequate means of urban

transportation. At the beginning of the industrial era, one could

walk from the periphery to the center of each of the largest

European and American cities in less than an hour. In the

1830s, the area occupied by each of the three largest cities in the

Western world—Moscow, London, and Paris—was less than 60

square kilometers. Today, by contrast, the built-up area of the

largest cities covers several thousand square kilometers each.

In large modern cities, mobility can be maintained only with an

elaborate system of transport, usually combining private and

public modes of travel. The frequency of face-to-face contact

among the millions of people living in large cities depends

entirely on the efficiency of a motorized urban transport system.

t h e s pat i a l pat t e r n o f l a b o r m o b i l i t y

Every day in urban areas, the millions of people who constitute

the active population leave their homes to travel to their places

of work, usually located in parts of the metropolitan area other

than the ones in which they live. Every evening these same people

come back home. In between, they may drop their children at

school, stop to buy groceries, or meet friends at a coffee shop.

These daily trips originate and terminate at people’s homes but

also include their workplaces and any number of amenities—

restaurants, museums, supermarkets, cinemas, etc. The

commute constitutes a daily tide moving back and forth in a

predictable manner, with peak hours and ebb times, from home

to workplace and amenities and back.

In addition to trips originated in residential areas by commuters

and consumers, economic activities generate freight trips

between firms’ locations and increasingly, with the growth of

e-commerce, from firms directly to their consumers in residential

locations. Firms in large cities need to be constantly supplied

both with merchandise to be sold in shops and with the materials

and parts to be used in manufacturing. These freight trips do

not follow the same patterns as commuting trips and are often

ignoredbyplanners.InatypicalcityinanOECD5 country,

freight trips may represent 10 to 15 percent of the total vehicle

kilometers traveled (VKT). When roads are congested, the tax on

productivity affects both labor and freight mobility.

A city’s economy is therefore dependent on the repetitive flow of

commuting and freight trips. If by chance a snowstorm, a flood,

or a public transport strike forces these trips to be canceled, the

city’s economy freezes immediately and remains frozen until the

daily commuting tide resumes.

5 The OECD (Organization for Economic Co-operation and Development) is a club of 34 rich countries with high

human development index committed to market economy and democracy


co m m u t i n g t i m e a n d co m m u t i n g co s t l i m i t t h e s ize o f

l a b o r m a r k e t s

Obviously,therearelimitstothemoneyandtimethatworkersare

willing to spend on commuting. These limits impose a constraint

on the commuting distance, and as a consequence, on the size

of the urban labor market. For very low-income workers—whose

income is nearly entirely devoted to food and shelter—the cost

of commuting is a more binding constraint than the time spent

commuting. As a household’s disposable income increases, the

cost of transport becomes a smaller fraction of income—typically

less than 15 percent, and the time spent commuting becomes

the major constraint for workers, limiting the size of the labor

market. Because the time spent on commuting is a dead loss

for both individuals and employers, the size and efficiency of a

labor market depends on how short, cheap, and comfortable the

commute is. The maximum cost in time and cash that workers

are willing to spend commuting will therefore dictate the size of

the labor market and, by extension, the productivity of a city.

Urban commuting surveys indicate that the median travel time

across cities and countries is, and for a long time has been,

remarkably stable, with an approximate mean of 30 minutes

eachway.Onlyasmallpercentageofcommutersinlargecities

have a total commuting time of more than one hour per day.

In 2009, the mean travel time in US metropolitan areas was 26

minutes; however, in New York, the largest US metropolitan area,

with 19 million people, it was 35 minutes. Figure 1 compares

the distribution of commuting travel time (one way) between an

average of US metropolitan areas and Gauteng, the South African

metropolitan area that includes Johannesburg and Pretoria

(12.3 million people in 2011). In spite of the difference in their

economies, urban structures, cultures, and topographies, the

majority of commuters, approximately a third in both Gauteng

and US cities, spend 15-29 minutes of travel time. The percentage

of commuters who spend less than 15 minutes of travel time

is significantly higher in the US cities, and the percentage

who spend either 30-59 minutes or more than 60 minutes is

significantly higher in Gauteng.

I consider an hour’s commute (one way) to be the absolute limit

when defining the spatial extent of a labor market. For a worker,

the number of jobs that can be reached within a travel time of less

than one hour defines the size of his labor market.

Figure 1: Distribution of commuting travel time in US cities and in Gauteng (South Africa)


Onecouldarguethatthenumberofjobsforwhichaparticular

worker would be qualified or in which he would be interested

would be much smaller than the total number of jobs accessible

within an hour of his home. This is true, but increasingly

specialized jobs have led to a greater dependence on the physical

proximity of people with other specialties and skills. In the

service industry in particular, which constitutes a large portion

of the jobs in large cities, large arrays of skills are needed in close

proximity. For instance, a lawyer who specializes in European

agriculture regulations would not be very productive if she were

surrounded only by people with the same skills. To be effective,

she will have to be in close contact with other specialists in

taxation and import tariffs, and she will need to engage the

services of workers who will fix her computer, clean her office,

deliver coffee to the board room and prepare and serve the

food that she will eat at lunch. In the same way, an unskilled

industrial worker is likely to work in a factory requiring a large

array of workers specialized in electronics, mechanics, labor law,

insurance and so on.

The idea that a lawyer needs to access only the area where lawyers

are likely to work while an industrial worker needs to access

only industrial areas no longer corresponds to the reality of job

distributioninalargemoderncity.OurEuropeanagriculture

regulations specialist may only be interested in a few jobs, and

these few jobs are likely to be randomly distributed among

many other jobs. For this reason, the larger the total number

of jobs, the greater the chances that a few very specialized jobs

will be among them. In addition, the larger the number of jobs

accessible within an hour’s commute, the better the ability to

change jobs when desired. This type of labor mobility – the ability

to change jobs within different economic sectors – benefits both

individual workers and the city economy by redistributing labor

where it will provide the most benefits.

t h e e f f ec t i v e s ize o f t h e l a b o r m a r k e t d e p e n d s o n

t r av e l s p e e d a n d t h e s pat i a l d i s t r i b u t i o n o f j o b s

The impact of travel speed, size of labor markets and jobs’

spatial distribution on urban productivity has been convincingly

demonstrated for European and Korean cities by Prud’homme

and Lee6 and for US cities by Melo, Graham, Levinston and

6 Prud’homme and Lee, 1998, “Size, Sprawl, Speed and the Efficiency of Cities”. L’OEIL, Observatoire de l’Économie

Aarabi.7 Prudhomme and Lee’s paper titled “Size, Sprawl, Speed

and the Efficiency of Cities” shows that productivity per worker

is closely correlated to the average number of jobs per worker

that are reachable in less than 60 minutes. In Korean cities, a 10%

increase in the number of jobs accessible per worker corresponds

to a 2.4% increase in workers’ productivity. Additionally, for

25 French cities, a 10% increase in average commuting speed,

all other things remaining constant, increases the size of the

labor market by 15 to 18%. In the US, Melo et al. show that the

productivity effect of accessibility, measured by an increase in

wages, is correlated to the number of jobs per worker accessible

within a 60-minute commuting range. The maximum impact

on wages is obtained when the number of jobs accessible

within 20 minutes increases; within this travel time, a doubling

in the number of jobs results in an increase in real wages of

6.5%. Beyond 20 minutes of travel time, worker productivity still

increases, but its rate decays and practically disappears beyond

60 minutes.

Both papers demonstrate that workers’ mobility – their ability to

reach a large number of potential jobs in as short a travel time

as possible, is a key factor in increasing the productivity of large

cities and the welfare of their workers. Large agglomerations

of workers do not insure a high productivity in the absence of

worker mobility. The time spent commuting should, therefore, be

a key indicator in assessing the way large cities are managed.

As Prud’homme writes in his paper: “[…] the benefits associated

with city size are only potential, they are contingent upon the

quality of management. City size would therefore define an

efficiency frontier, with effective efficiency often significantly

below this frontier.” The “quality of management” as defined by

Prud’homme is in a large part the ability of the local government

to adapt the transport system to the spatial structure so that

workers can access a maximum number of jobs in less than 60

minutes of travel time.

The effective size of a city’s labor market is, therefore, not

necessarily equal to the number of jobs available within a

metropolitan area but to the average number of jobs per worker

et des Institutions Locales, IUP — Université de Paris XII

7 Patricia Melo, Graham, Levinson and Aarabi, 2013, “Agglomeration, Accessibility, and Productivity: Evidence for

Urbanized Areas in the US” paper submitted to the Transportation Research Board, Washington DC.


accessible within a one hour commute. Depending on the speed

of the transport system, the effective size of the labor market

could be equal to the total number of jobs available in a

city or to only a fraction of it. The location of workers’ residences

relative to their jobs and the speed of transport will determine

the effective size of a labor market and, therefore, the additional

productivity that could be gained by the scale and agglomeration

economies described earlier.

I will illustrate the relationship between speed of transport,

effective size of labor market and spatial distribution of jobs by

using a schematic representation of a city as shown on Figure 2

Imagine a linear city where workers’ residences are spread evenly

between a and e. Jobs are concentrated in only three locations

b, c and d. Each location contains 1/3 of all jobs. The speed of

transport is uniform within the city and is represented by the

arrows showing travel time between different points. It takes

two hours to travel from a to e, on opposite outer edges of the

hypothetical city.

Workers living between b and d can reach 100% of the jobs in less

than one hour, but workers living between a and b can reach only

the jobs located in b and c in less than one hour; jobs located in

d are out of reach for workers living between a and b. Similarly,

workers living between d and e can reach only the jobs located in

c and d; the jobs located in b are out of reach. As a consequence,

50% of the workers (those living between b and d) have access

to 100% of the jobs in less than 1 hour of travel time while the

other 50% (those between a and b and between d and e) only have

access to 2/3 of all the jobs. Therefore, the effective size of the

labor market represented in Figure 2 is only 83% of all the jobs

available in the city: 50% of 3/3 + 50% of 2/3 = 83.3%. If the speed

of transport could be increased so that one could travel from a

to d and from e to b within one hour, rather than the 90 minutes

each trip currently takes, then the effective size of the job market

would be 100% of all jobs available (100% of 3/3 = 100%).

In a less schematic city, the effective size of labor markets can be

calculated as follows: Let us assume that the city is divided into

polygons identified by their number i; then, the effective size of

its labor market can be expressed by:

J =Σ(wi ji)/Σni

where:

J is an indicator of the effective size of the labor market expressed

as the average % of total jobs accessible in less than one hour per

worker;

wi is the number of workers living in location i;

ji is the number of jobs accessible within one hour travel time of

location i;

ni is the number of jobs in location i;

This type of calculation would have been prohibitively labor

intensive before the availability of GIS technology, but it is now

quite feasible to update this indicator regularly. Different

transport modes and networks could be tested for their potential

impact on the effective size of the labor market.

The effective size of a labor market depends on commuting

travel speeds and the relative location of workers’ residences to

their jobs. This dependence may be illustrated in a less abstract

way by representing a city as a two dimensional object, rather

than the one dimensional, linear representation of Figure 2, and

by showing alternative arrangements for travel speeds and job

locations.

Figure 3 shows a schematic representation of an urban built-

up area, represented by a circle. Within this circle, smaller red

circles represent job locations. Horizontally, I have shown three

types of spatial distribution for jobs: monocentric, where all jobs

are concentrated in a central business district (CBD); polycentric,


Figure 2: Distribution of workers residence and job location in a hypothetical linear city

where jobs are concentrated within three clusters; and dispersed,

where jobs are uniformly distributed within the built-up area. For

each pattern of job distribution, an arrow shows the maximum

travel distance that a worker can cover in one hour from the outer

edge of the urban area. The different arrows’ lengths correspond

to different travel speed.

We will see how different commuting speeds have an impact on

the effective size of the labor market depending on the spatial

distribution of jobs.

In the three graphs on the top row of Figure 3, I have adjusted

the speed of travel to allow for accessibility to all jobs in less than

onehour.Onthebottomrow,atalowertravelspeed,theworkers

residing on the periphery can access only a fraction of the jobs in

less than one hour. The labor market in this case is fragmented

and is therefore less efficient than the unified one represented

in the top row. Workers who live in a more central area may have

access to all jobs within the built-up area, but workers living

on the periphery have access to only a fraction of the total jobs

available in the city. In this case, then, the implied productivity

of a large labor market is not fully realized. A decrease in

commuting travel speed fragments large labor markets into

smaller ones and results in a decrease in urban productivity.

Increasing travel speed decreases the difference between the

effective labor market (the number of jobs accessible within an

hour commute) and the nominal labor market (the total number

of jobs in a metropolitan area.)

For a given built-up area, the pattern of job distribution is

important in defining access to the labor market. When jobs are

clustered in a CBD, the distance from all jobs to all residential

locations is much shorter than it is when jobs are randomly

distributed within the built-up area. This does not necessarily

demonstrate that the CBD model is the most efficient pattern or


Figure 3: Labor markets, speed, and job location

that it will ensure full access to the labor market for everyone.

It is true that a centrally located CBD8, containing one hundred

percent of all jobs, would decrease the distance from one’s

residence to one’s job for everyone. However, the size of labor

markets is limited not only by distance but also by travel time.

Therefore, speed of transport (distance/time) is the key parameter

in allowing access to the maximum number of jobs.

The convergence of all commuting routes9 toward a CBD usually

creates congestion and slows the speed of travel. In contrast,

when jobs are dispersed in suburban locations, there is no

convergence of routes, and transport speed is usually faster. The

average surface transport speed in the center of Paris (within

5km of City Hall) is about 12 km/h; rush-hour speeds in the

suburbs (20 km from City Hall) are about 50 km/h. Thanks

to improvement in GPS technology, we are now able to check

variations in rush hour speeds on the internet in real time for

many cities in the world.

In areas where the major road network was originally designed

for a monocentric city, commuting routes from suburb to suburb

may be less direct than they should be; this is the case in Paris,

Atlanta and Shanghai. Initially, suburb to suburb commuting

routes may have to follow minor roads and may include awkward

major road crossings. There is usually a long time lag before a

municipality is able to adjust the design of a major road network

from monocentric to a grid-like pattern that will better serve new,

emerging routes among suburbs.

It is outside the scope of this paper to further discuss the

influence of different transport modes – cars, public buses, or

subways – on transport costs and travel time for various types of

urban spatial structures in which population and job densities

are distributed differently within the built-up area. The schematic

representation of spatial structure in Figure 3 is very crude, but

it clearly demonstrates the impact of speed of travel and job

location on the effective size of a labor market.

8 Not all CBDs are located at the center of the built-up area. Mumbai’s CBD, for instance is located at the Southern

tip of a peninsula while the centroid of the built up area is located more than 15 km to the north. This situation is

relatively rare, as market forces tend to “re-center” the CBD toward the center of gravity of a city’s population.

9 The term “commuting routes” defines an itinerary from one place to another, which may have to follow minor roads

in the absence of major roads linking the point of origin to the desired destination. Commuting routes are therefore

independent from the existing design of major roads which may often converge toward a central point.

In cities where job accessibility has been measured, the number

of jobs accessible by commuting time shows large variations. For

instance, Prud’homme noted that in Seoul, “in 1998 the average

worker has in 60 minutes access to only 51% of all the jobs

offered by the city; and the average enterprise has 56% of all the

workers at less than 60 minutes.” The additional subway lines

built since that date must have increased the effective size of the

labor market in Seoul. A comparison of car commuting across

US cities in 2010 calculated by David Livingston10 shows amazing

differences in accessibility between US cities (Figure 4). Within a

30 minute drive, 2.4 million jobs can be accessed in Los Angeles

compared with 0.6 million in Atlanta. However, 4 of the 5 cities

represented in figure 4, allow access to all jobs within 60 minutes.

t h e l a b o r m a r k e t s h a p e s t h e pat t e r n o f co m m u t i n g

t r i p s

As we have seen in Figure 3, a majority of jobs may be

concentrated in a central business district or be clustered in

several centers or be completely dispersed across a metropolitan

area. Next, we will look only at the possible trip patterns that

would allow the labor market to function within each of the

following spatial distributions of jobs.

Figure 5 illustrates in a schematic manner the most usual trip

patterns in metropolitan areas depending on the concentration

or dispersion of jobs. There are three observed models of

commuting route patterns, labeled from A to C on Figure 5:

A. The monocentric model– most jobs are

concentrated in a dense Central Business

District (CBD); trip routes follow radial roads and

convergetowardtheCBD.Ofcourse,norealcity

is ever strictly monocentric; a number of jobs are

necessarily found inside residential areas, for

instance in schools, dispensaries, gas stations and

grocery stores.

Monocentricity is really measured by degree rather

than in absolute terms. A city where more than 50%

of the jobs are located in the CBD is dominantly

monocentric. To my knowledge, no metropolitan

area with a population above 5 million meets this

10 David Levinson, “Access Across America,” 2013, Center for transportation studies, University of Minnesota



Figure 5: Possible trip patterns in metropolitan areas

Figure 4: Average number of jobs accessible by workers in various US cities

criterion.

B. The polycentric or dispersed model – most jobs

are concentrated in small clusters or completely

dispersed among residential areas; trip routes are

randomly distributed within the built-up area. If

speed of transport allows it, some workers will

commute from one edge of the metropolitan

area to its opposite edge. As in the monocentric

model, workers residing closer to the centroid of

the built-up area are closer to all the jobs than are

workers residing at the edge. Firms located closer

to the built-up area centroid are also closer to all

workers. However, because commuting speeds

are usually not the same near the centroid and at

the periphery, firms located close to the periphery

might be accessible in shorter time by more workers

than firms located close to the centroid. In part,

this explains why firms do not cluster close to the

centroid even though doing so would put them at a

shorter distance from all their potential workers.

C. The composite model – a significant fraction of all

jobs are concentrated (say, 30% for instance) in a

dense CBD, but the majority of jobs are randomly

distributed in the rest of the built-up area. Trip

routes toward the CBD follow radial roads while

trip routes toward dispersed jobs are randomly

distributed but usually avoid the congestion of the

CBD. This is the most usual pattern of trips in large

cities in Asia and Europe.

There is also a fourth model of trips that doesn’t exist in the

real world but is very often presented in master plans as being a

desirable alternative to the three trips’ patterns described above.

Because of the prevalence of this conceit in many urban master

plans, this utopian alternative trip pattern, labeled D in Figure 5,

needs to be discussed:

D. The so-called “Urban Village Model” – jobs are

concentrated in many small clusters. In this model,

there are many centers, but commuters travel

only to the center that is closest to their residence.

The trips toward each job cluster follow radial

routes centered on each cluster and behave as if

each cluster were an isolated, monocentric city.

According to this model, a large city can be made up

of a number of self-sufficient, small monocentric

cities.

Unfortunately, the urban village model exists only in the mind of

urbanplanners.Otherwise,itwouldbeaveryattractivemodel,

which is why urban planners favor it, as it would not require

significant investment in transportation or roads. Furthermore,

it would dramatically reduce vehicle kilometers travelled (VKT)

and, as a result, greenhouse gas (GHG) emissions. According to

the proponents of this model, everybody could walk or bicycle to

work, even in a very large metropolis. To allow a city to grow, it

would only be necessary to add more clusters. The assumption

behind this model is either that urban planners would be able to

perfectly match work places and residences, or that workers and

employers would spontaneously organize themselves into the

appropriate clusters.

This model does not exist in the real world because it contradicts

the economic justification of large cities: the efficiency of large

labor markets. Employers do not select their employees based on

their places of residence; neither do specialized workers select

their jobs based on proximity from their residences. However,

as I have postulated above, employees may be reluctant to

accept a job located beyond a one-hour commuting time. This

commuting time limit would not allow the creation of urban

villages.

The “urban village model” implies a systematic fragmentation

of labor markets within a large metropolis and does not make

economic sense in the real world. A firm that would be satisfied

to restrict the selection of its employees to the vicinity of its

factory or office would not need to locate in a large metropolis

where rents and salaries are higher. This firm could locate in a

small town where the unspecialized workers it seeks could be

recruited for a lower salary. In the same way, a worker living in

a large city and looking for a new job would try to maximize job

satisfaction, measured in part through salary, level of interest

in the work and its compatibility with skillset, attractiveness

of the work environment, etc. The time spent commuting


might certainly be a consideration in seeking a job, but if the

commuting time were less than one hour, it would likely not be a

determining one.

The five satellite towns built around Seoul are an example

of an attempt to implement the urban village concept. The

government built the new towns under the assumptions that

they would be self-contained and that most inhabitants would

work and live within their own towns. To achieve this objective,

planners carefully balanced the number of projected jobs in

each town with the number of projected inhabitants. However,

subsequent surveys showed that most people living in the new,

satellite towns commuted to work in the Seoul metropolitan area,

and most of the jobs in the satellite towns were filled by people

living outside of them. The trip pattern found in satellite towns

is consistent with the hypothesis made at the beginning of this

paper: a large unified labor market is the justification for large

cities. It’s likely that some households initially decided to move

to Seoul’s satellite towns because apartments were cheaper than

in Seoul’s core city or because the environment was better and

newer. It is also probable that when these households moved,

the heads of household were already employed somewhere in

Seoul. After all, had they not been employed, they likely would

not have been able to buy a new apartment. Furthermore, after

moving to a satellite town, it is unlikely that they would quit their

current jobs to find equivalent, vacant jobs within the town. The

same reasoning could be made for firms moving to a satellite

town. A firm might move from the central city to find cheaper

rents or more space, but many of its employees would likely

decide to keep their jobs and commute from the core city to the

satellite town.

h ow co m m o n a r e e ac h o f t h e t h r e e s pat i a l

d i s t r i b u t i o n m o d e l s?

The spatial distribution of jobs, and as a consequence

commuting trip patterns, evolve as cities become larger and

more affluent. The monocentric model is a simple, primitive

city model that inevitably evolves over time into a more complex

form,morecloselyresemblingthecompositemodel.Oncejobs

have dispersed into a pattern similar to the dispersed/polycentric

model, it is unlikely that they will eventually concentrate again

into a dense, central CBD.

This path dependency11 rule, common to all evolving shapes, is

a reality that should seriously limit the freedom of planners to

dream up new urban forms. Planners should take into account

the path dependency of city shape when designing for new

transport systems.

None of the three models discussed above are immutable. Future

urban labor markets, for instance, might not require as many

face-to-face interactions between employees, customers and

suppliers as they have in the past. New models of trip patterns

might emerge in the future reflecting the new requirements of

an evolving labor market. For instance, the recent emergence

of telecommuting has put into question not only the pattern of

commuting trips but also the very need for commuting. As such,

we should remain agnostic regarding the patterns of commuting

trips twenty years from now. However, we can look at the trend in

trip patterns over the last decades to inform our understanding

going forward. This trend reflects path dependency, mentioned

above.

What effect could a large increase in telecommuting have on

current trip patterns? So far, the effect has been modest. In fact,

there is a hint that this modest trend may reverse itself among

high tech companies that were the first to initiate it. In 2013,

Yahoo’snewCEOannouncedareversalofitstelecommuting

policy, which confirmed what we already knew: that serendipitous

face-to-face contact between professionals is necessary for

innovation.

However, the question remains: how often should those face-

to-faceinteractionsoccur?Onceaweek?Everyotherday?

How large should the groups needing face-to-face interaction

be? How much serendipity is required to generate innovation?

Whatever the answer, telecommuting will certainly decrease

daily commuting flow and change traffic flow, but it will not

completely eliminate a worker’s need for spatial proximity to

his/her employer or to other workers with complementary skills.

It is quite possible that telecommuting will decrease in firms

requiring innovation and increase in firms engaged in routine

11 Path dependency refers to situations in which options taken in the past limit the number of options available in the

future. The concept is commonly used in history, evolutionary biology and economics but is obviously applicable to

urban development. For instance, in evolutionary biology a group of primitive living cells could possibly evolve in into

a mammal or into a fish. But once the cells have evolved into a fish, they cannot possibly evolve into a mammal and

vice-versa.


dataprocessing.Onelessonisclear,wecannotplanforit,butwe

must monitor carefully the spatial implication of this trend and

support it with adequate transport infrastructure.

Although the large metropolises of the world show a great

variety of histories, cultures and incomes, the trends, when the

data exist to measure them, seem to converge toward a more

spatial dispersion of jobs. This trend seems counterintuitive,

particularly as the CBDs in an increasing number of cities

compete for the distinction of world’s tallest skyscraper. But

we must realize that a prime office skyscraper contains fewer

workers per hectare than the five-story sweatshop that it likely

replaced.

Because of the very low built-up densities of US cities, their

spatial trend might not be representative of most world

cities. However, the US trend has the advantage of being well

documented and may still provide some insights into the way

changes in labor markets impact urban land use. In 1995

and 2005, Alan Pisarski conducted the most comprehensive

nationwide study of commuting in the US. The trends he

measured over a ten-year interval are clear: the ratio between

jobs and resident workers is decreasing in central cities and

increasing in suburban areas. Pisarski’s studies clearly show

that, on average, US metropolitan areas are slowly evolving from a

composite model (Figure 5, pattern C), to a more dispersed model

(pattern B). Furthermore, Pisarski’s studies indicated that the

job concentration in the traditional CBD is constantly decreasing

not in absolute terms but as a proportion of the total number of

metropolitan jobs.

Pisarski’s reports show that in smaller US metropolitan areas,

those below 100,000 people, commuting trips12 to the CBD

represent about 50% of all trips – a good approximation of the

monocentric model. However, for larger metropolitan areas,

those with population above two million people, the trips to

the CBD drop to about or below 24%. This is also observed in

very large metropolises with a well-marked, dominant CBD

such as Seoul, New York or Paris. For instance, in the New

York metropolitan area, 24.3% of trips are from the suburbs

12 Commuting trips include only trips between a residence and a work place. Other trips, for shopping, social life or

leisure, are counted separately. Commuting trips are often a fraction of all trips, but they are the most important for

proper functioning of the labor market. In addition, most commuting trips take place at rush hour and therefore test

the capacity limit of the transport system.

to Manhattan or within Manhattan; 2.1 % of trips are from

Manhattan to the suburbs, and the large majority, 73.6%, are

from suburb to suburb. The pattern of trips in metropolitan New

York illustrates what I have called the composite model (Figure 5,

pattern C).

OutsidetheUS,thetrendsoflargemetropolitanareasalsoseem

to move toward the composite model, even in cities like Paris,

with a historically dominant and prestigious center and a radio-

concentric transit system providing excellent access to the center.

In Paris’ metropolitan area (defined as the Ile de France region),

trips within and to the Paris municipality (historical Paris)

represent 30% of the total commuting trips, and 70% of trips are

from suburb to suburb (Figure 6).

Large metropolitan areas of Asia, although much denser than

US or European cities, are showing the same trends toward the

suburbanization of jobs and people. Seoul’s metropolitan area,

with a population of 24.7 million in 2010, is representative of the

trend in prosperous East Asian cities that have seen a significant

increase in population and household income in the last 30 years.

In Seoul, between 2000 and 2010, the population decreased by

0.5% in the central city13, and it increased by 92% in outer suburbs

located more than 20 km from the city center (Table 1). During

that same ten year period, the spatial distribution of jobs has

been less dispersed, with 16% of the new jobs being added to the

CBD, while 59% were added to outer suburbs more than 20 km

from the CBD. Seoul, however, still remains more monocentric

than Paris or New York, with 31% of the total metropolitan jobs

concentrated in the central city.

As we can see from the historical trend the monocentric model

tends to break down when a city becomes larger. However,

empirical evidence does not show an obvious population size

threshold beyond which cities cease to thrive as dominantly

monocentric. Sometimes, topography – rivers or mountains

– prevents direct communication from suburb to suburb and

therefore maintain monocentricity in spite of a large population.

An original network of primary radial roads would reinforce a

high degree of monocentricity as a city expands by making it

easier to go to the CBD than to peripheral locations. This type

13 In this case, I defined Seoul central city as the area within a circle of 10 km radius centered on Seoul City Hall. This

area includes three distinct CBD-like areas with an intense spatial concentration of jobs.


of radial network can be seen in European cities like Berlin,

Copenhagen or Paris. By contrast, a primary grid network

would rapidly encourage the creation of sub centers with good

overall accessibility as a city develops. This has been the case

forLosAngeles,Houston,andOmaha.Thegridsinthesecities

sometimes become irregular, but the availability of wide roads,

perpendicular to the radial roads at the fringe of urbanization,

stimulates the creation of sub-centers, and perhaps even of job

dispersion, because wide roads allow for higher driving speeds

and, therefore, for faster accessibility to areas farther away from

the radial roads. A grid network of roads, therefore, encourages

an early shift toward polycentricity.

THE AFFORDABILITY OF LAND AND FLOOR

SPACE ALLOWS ALL INCOME GROUPS TO

PARTICIPATE IN THE LABOR MARKET

t h e e f f i c i e n t o p e r at i o n o f l a b o r m a r k e t s r eq u i r e s

m o b i l i t y

Mobility should be understood in two ways: first, it is the

ability to move quickly and easily between locations within a

metropolitan area; second, it’s the ability to locate one’s house or

one’s firm in any location within a metropolitan area. Mobility

is, therefore, not only the ability to move rapidly between origin

and destination but also the ability to change the origin and

destination of commuting trips as circumstances dictate.

The ability of households and firms to choose where they locate

depends upon the availability of land and floor space in the areas

of the city they deem to have the most favorable benefits— for

households, the proximity to jobs and amenities, and for firms,

the proximity to clients, employees and suppliers. Theoretically,

the lowest-income household or the least well-capitalized firm

should be able to locate anywhere in a city, even where land


Figure 6: Trip Patterns in Metropolitan Paris


Figure 7: Seoul - Changes in population and job distribution between 2000 and 2010

Table 1: Seoul - Spatial changes in the distribution of population and jobs 2000-2010

is the most expensive, if they are allowed to consume as little

land as they require. The food carts of Manhattan and the tiny

paanwalas14 stalls of Mumbai demonstrate this point. These

small businesses sell cheap products but thrive in their expensive

locations by consuming only two or three square meters of land,

as opposed to the several hundred square meters occupied by

most shops in the same area. Farmers’ markets and flea markets

are other examples of the trade-off between location and land

consumption allowing low-margin businesses to thrive in

expensive locations.

Equivalent examples exist for housing; for instance, Paris’

chambres de bonne15 are located in the city’s most expensive

neighborhoods. Typically as small as nine square meters, they

allow students or low-income workers to afford housing in a

very favorable location. The residential plots in Indonesia’s

kampongs16 are another example of the demand for centrally

located residences with low land and floor consumption. The

availability of these residences allows the poor to decide about

the trade-offs they want to make regarding location and land

consumption. If water, sanitation, and refuse management are

adequate, a small, inner city location might be a desirable trade-

off compared to a suburban one far from jobs, amenities, and

social services. Whether “Chambres de bonnes” or “Kampongs,”

the neighborhoods in which such dwellings are located are by no

means slums, in spite of the very small size of their dwellings.

Unfortunately, well-intentioned regulations often prevent

the poor from making these trade-offs between floor space

consumption and location. Urban regulations typically require

a “generous” minimum floor space standard for housing. These

well-intentioned regulations exclude the poor because the high

price of floor area makes it too expensive for them to afford the

minimum standard. Chambres de bonne exist only in housing

14 Paanwalas sell paan, a mixture of betel leaf, areca nuts, and tobacco. They are small, informal, thriving retail

businesses in the streets of Indian cities.

15 “Chambres de bonnes” or maids’ rooms were independent rooms of about 10 to 12 square meters built under the

roofs of opulent buildings in Paris and provincial towns, usually with common bathroom facilities on the same floor.

When the households in these buildings could not afford maids anymore, the rooms became the cheapest rental rooms

on the market. Their low cost has been maintained over the years in spite of their excellent location because they are

located on the 5th or 6th floors of buildings without elevators.

16 A Kampong, which means village in Bahasa Indonesian, is an informal but legal residential neighborhood in

Indonesian cities. The lot sizes vary by income with the smallest being around 10 m2. Streets in a Kampong are

usually 2 or 3 meters wide; some passages between houses not wider than one-half meter.

built before 1930; they are prohibited in apartment buildings

that are more recent. These unfortunate regulations reduce the

mobility of the poor. As a consequence, their participation in the

full labor market is also restricted.

In South Africa, the government housing program for the poor

illustrates how well-intentioned planners may limit residents’

mobility, reduce their participation in the labor market, and

increase their travel time. Starting in 1995, the government

embarked on a massive housing program to improve the living

conditions of the victims of apartheid. The program aimed to

provide subsidized housing to about 80 percent of the poorest

South Africans. By 2012, it had already delivered 3.5 million

urban dwelling units, a unique quantitative achievement for

a government housing program. The standards are generous:

400 square meters per lot, 65 square meters of floor space per

house, wide vehicular access roads, and schools equipped with

large sports grounds and within walking distance, etc. The

space standards are fixed and similar all over urban South Africa.

The subsidy is also fixed, amounting to about 150,000 rand (US

$16,000) per dwelling in 2012. The only dependent variable is

the price of land. Land for this massive program must be very

cheap to allow beneficiaries to enjoy the high space standards

prescribed by the program. As a result, the new subsidized

housing projects are all located in the far periphery of cities, in

settlements that are too dispersed to be easily serviced by public

transport. Because of the remote location, the beneficiaries of

these projects have a hard time finding employment. Those

who are employed pay as much as 50 percent of their income on

transport, even when sharing taxis with other commuters. This

dramatic example illustrates why planners should not make

trade-off decisions involving location and land consumption for

households and firms.

In the case of South Africa, the outcome is indeed comfortable

housing for poor people, but the large subsidy attached to a house

in a distant location prevents the beneficiaries from participating

in the labor market. High-standard housing provided to poor

people in a remote location becomes a poverty trap. The

subsidies are not to blame here. The error is not only tying the

subsidy to a specific location but also deciding on the location

verse land consumption trade-off without having solicited input

from the beneficiaries. A portable subsidy—that is, a lump sum


given to poor households to use for shelter anywhere in the

metropolitan area—is significantly preferable.

Because planners lack information to make informed decisions

about the difficult trade-offs between location and land

consumption, they shouldn’t have the authority to make these

typesofdecisions.Onlyafreemarketallowshouseholdsand

firms to choose the trades- offs that best allow them to maximize

their comfort and their participation in the labor market. To

benefit fully from a large labor market, households and firms

must have ease of mobility. As such, there must be sufficient,

affordable options that allow them to make the trade-off

decisions between location and land consumption that best suit

their needs.

CITIES VIEWED AS LABOR MARKETS:

OPERATIONAL IMPLICATIONS

Looking at cities as unified labor markets should change the

generally negative views that urban planners hold regarding

mobility. Planners believe that one of their most important tasks

is to decrease nuisances created by urbanization and particularly

by vehicular traffic. They are right; decreasing congestion and

pollution is one of the most pressing challenges brought on by

urbanization. However, planners, in their enthusiasm to reduce

nuisances, often fail to understand the differences between the

objective of increasing mobility and the constraint of decreasing

the nuisances it creates.

Oneofthemainobjectivesofurbanplanningismaintaining

mobility – preventing an increase in commuting time as the size

of the labor market increases. In other words, the main objective

of planning should be to increase the speed of urban transport as

a city’s size increases. Decreasing the level of nuisances due to

transport is a constraint that may increase the cost of achieving

the objective. But this cost is fully justified if the marginal

cost of reducing nuisance is lower than the marginal increase

in productivity due to the expansion of the labor market. For

instance, charging vehicles in proportion to the pollution they

create, or imposing tolls on highways to reduce congestion would

increase the cost of transport, but at the same time would also

increase its efficiency, allow for higher speeds and improve

environmental quality.

However, too often planners substitute the constraint for the

objective. For instance, advocates of “smart growth” imply that

the reduction of congestion and pollution is the main objective

of urban planning. They soon realize that nothing would reduce

congestion and pollution more certainly than a decrease in

mobility. Reducing mobility is then considered a desirable

outcome and is the logical consequence of the confusion between

objectives and constraints. For this reason, planners design

“urban village” spatial arrangements (Figure 5, pattern D), which

implicitly reduce mobility.

Matching employment location with residence in large

metropolitan areas has become the Holy Grail of urban planners.

This recurrent conceit, motivating many master plans and many

land use regulations17, can only be explained by ignorance of the

economic efficiency of large labor markets and the mobility that

large labor markets require in order to function. The choice of

spatial location is best left to households and firms themselves.

The functioning of labor markets is my guiding principle in

evaluating alternative spatial arrangements. “Mobility” is the

ability to reach any area of a metropolitan area in as short a travel

time as possible, and “affordability” is the ability of households

and firms to locate in whichever area they deem will maximize

their welfare. Increasing mobility and affordability are the two

main objectives of urban planning. These two objectives are

directly related to the overall goal of maximizing the size of a

city’s labor market, and therefore, its economic prosperity.

17 For instance, Stockholm urban regulations require a job housing balance in each neighborhood, in spite of statis-

tics showing that when this balance is achieved it doesn’t decrease trip length. Allowing mixed land use where firms

and housing can be found in the same location is an excellent thing but mandating it is illusory and only slow down

the development process. In addition, the number of jobs per commercial structure could vary a lot in time.


Documents

CITIES AS LABOR MARKETS - Marron Institute · 2019-07-11 · CITIES AS LABOR MARKETS + ALAIN BERTAUD ABSTRACT A city’s welfare depends on its labor market. As long as a labor market